Contents9
- 01Shopify or an open source platform?
- 02What makes a product page sell?
- 03Can AI write product pages?
- 04How do you feed your catalogue to Google Shopping and marketplaces?
- 05How do you keep stock and payments under control?
- 06Which numbers should you track each week?
- 07When should you open a new market?
- 08Where to start
- 09Frequently asked questions
Opening an online store has never been easier: a few hours are enough to put products on sale. Making it pay is another matter. Stores that plateau rarely have a platform problem: their product pages do not answer buyers’ questions, their catalogue differs on every channel, or nobody reads the numbers that would say where to act.
This guide opens our Ecommerce section. It follows the order in which a store is built and turned around: the platform, the product pages, the catalogue and its channels, stock and payments, the numbers, then growth. Each section links to the article that goes deeper. If your store needs to be built or put back on track, our ecommerce offer starts from your real sales, not from a theme.
In short
For a standard sale, a hosted platform like Shopify is usually enough. An open source solution makes sense when the sale falls outside the standard.
A product page sells when it answers, without searching, the total price, the delivery time, returns and what real customers say.
One clean catalogue feeds the store, Google Shopping and marketplaces.
Stock has a single source of truth that deducts every sale whatever the channel.
Five numbers read each week tell you whether to act on traffic, the product page, the offer or loyalty.
Shopify or an open source platform?
For a small business with a standard buying journey (a catalogue, a cart, a payment, a delivery), Shopify is usually enough: the subscription covers hosting, security and updates. An open source platform like Medusa makes sense when the sale falls outside that frame (prices specific to each client, an ERP that drives stock) and a team can keep it alive. Compare over three years, development included: we detail it in Shopify or Medusa: which one for a small business?.
On Shopify, the cost that drifts fastest is not the subscription: it is the apps added one by one, each with its price, its code in the theme and its effect on speed. A yearly inventory is enough to sort them out, as we explain in Shopify: when one more app costs too much. And if you decide to leave, the migration is prepared URL by URL: Leaving Shopify without losing your SEO.
In B2B the question is different: each client sees their prices and terms after logging in, and a quote request becomes an order with no re-entry. See B2B store: client pricing, quotes and reorders.
What makes a product page sell?
A page that sells answers the questions buyers ask before paying: what exactly the product is, the total price with delivery, the delivery time, the return terms, and what real customers say. Those answers reassure more than a description that praises the product. The detail, element by element, is in A product page that converts.
What reassures the buyer
The total price, delivery included, visible straight away
A dated delivery estimate
Return terms in one sentence
Verified reviews, including the less glowing ones
What makes them leave
Fees discovered at checkout
A description that praises without describing
Missing or contradictory specifications
A slow page on mobile
Reviews count twice: they reassure buyers and, marked up as structured data, they help search. How to ask every buyer, not just the happy ones, is explained in Customer reviews: collect, display and rank. And speed is not a technical detail: a Deloitte study for Google linked it to conversions on mobile retail sites, as we report in Store speed: what a second costs in sales.
Can AI write product pages?
It can help write thousands of them, on one condition: never guess. First define the list of attributes for each product family, let AI extract them from existing texts, and measure or ask the supplier for whatever is missing. An invented value on a product page is a false promise made to a customer. The method is in Product descriptions and AI: writing without inventing, and its large-scale version, with automatic checks and batch review, in A catalogue of thousands of products: AI under control.
How do you feed your catalogue to Google Shopping and marketplaces?
Google Shopping, Amazon and other marketplaces rest on the same base: a clean catalogue where every product has an accurate identifier, title, description, image, availability and price, plus its GTIN when it has one. One catalogue, fed to every channel, saves fixing the same error in five places. See Google Shopping and marketplaces: preparing your store.
Once the feed is clean, the campaign is won on titles, prices and excluded queries: Google Shopping: what makes a campaign rise or sink. And before opening on Amazon, count the margin after subscription, selling fees and logistics: Amazon alongside your store: margin, stock, dependence.
One catalogue, several channels
- The source
A single catalogue, in the store, the ERP or a dedicated tool.
- The check
Identifiers, prices and availability verified before sending.
- The feed
One feed per channel, store, Google Shopping, marketplaces.
- The return
Every sale deducts the shared stock.
How do you keep stock and payments under control?
Selling the last item twice is the mistake that costs the most trust. It is avoided with a single source of truth for stock (the store, the ERP or the till) that deducts every sale whatever the channel, a safety margin on the last units and an alert when figures diverge. See Inventory synced across store, marketplaces and shop.
On payments, the real fees combine a percentage and a fixed part that vary by card, country and currency, plus disputes: we compare them in Payments: the real fees of Stripe, PayPal and Shopify Payments. And returns cost far more than the refund: Product returns: a hidden cost to manage, not endure.
Which numbers should you track each week?
Five are enough, read together and compared with the previous week: visits by source, conversion rate, average order value, share of returning customers and margin after acquisition costs. Each points to a different lever: traffic, product page, offer or loyalty. The table, and how to read it, are in Online store: five numbers to track every week.
Two levers are set once and pay back every week: automatic emails (welcome, post-purchase, repeat purchase), covered in Store emails, and the abandoned cart reminder, which only sells if it answers the reason for abandoning: Abandoned cart: reminders that sell without nagging. Finally, the store’s internal search tells you each week what customers look for without finding it: Store search: when customers can’t find it.
From the field
We are putting a Shopify plush toy store back on track: product pages, Google Shopping feed, apps installed over the years. The first job was not to add anything, but to list what was in place and what each element brought in.
When should you open a new market?
When the first one runs, and based on numbers: visits and orders already coming from the target country, delivery cost, VAT, customer service in the language. Translate first the best-selling product pages and the pages that commit you (delivery, returns, terms), show prices in the local currency, then measure. See Multilingual, multi-currency store: where to start, and our Website translation section for the translation method.
A newer question also arises: should AI agents be allowed to read your catalogue, or even buy for their users? It is a commercial choice, action by action, which we examine in Websites are shutting out AI agents: what about yours?.
Where to start
- Read your five numbers.
They tell you whether the problem is traffic, product page, offer or loyalty.
- Rework the best sellers’ pages.
Total price, delivery time, returns and reviews visible without searching.
- Clean the catalogue at the source.
One place is the reference, the channels receive it.
- Audit your apps.
Keep those that bring in more than they cost, speed included.
- Automate the recurring emails.
Welcome, post-purchase, repeat purchase, then the abandoned cart.
If you want us to do it with you, our ecommerce offer builds or turns around your store and connects it to your channels. For a first question, write to us.
Frequently asked questions
Is Shopify suitable for a B2B small business?
It is for a simple sale, and its B2B features cover client-specific prices and payment terms depending on the plan. As soon as prices, stock and orders must come from an ERP, the quality of the connection to that ERP matters more than the platform itself.
Should you be on Amazon as well as your own store?
Only if the margin stays positive after subscription, selling fees and logistics, and if stock stays synced between the two channels. Amazon brings volume but keeps the customer relationship: your own store remains where you build loyalty.
How long does it take to turn around a store that has plateaued?
The first effects often come from simple fixes on the best sellers’ pages, fees shown too late and useless apps. The deeper work, a clean catalogue and connected channels, takes longer. Measure each change on your five weekly numbers rather than by feel.
Can AI run a store for me?
It can prepare product pages, sort customer messages or comment on the week’s numbers. It does not replace the decisions that commit the store: prices, the offer, the answer to a dispute. And everything it writes about a product must be verifiable in a source.
Jim · founder of Brain Plus AI
Also in e-commerce

Choose and price6 min
Shopify or Medusa: which to choose for an SME?
Shopify charges a subscription and sometimes a commission; an open source platform like Medusa costs nothing in licence fees but needs a team to develop and run it. Here is how to compare the two over three years, and when each one wins.

Practical guide5 min
Product page that converts: what buyers look for
Before paying, buyers check four things: what exactly the product is, what they will pay in total, when they will receive it and what happens if it does not suit them. A page that answers those questions sells better than a page full of fine phrases.

Practical guide8 min
Product descriptions and AI: writing without making anything up
An AI can write hundreds of product descriptions overnight, and slip in a size, a material or a promise nobody has checked. The order of work that avoids this trap, what Google asks of a description, and what must stay in the hands of a reviewer.
